trunk/proteinstructure/Graph.java

package proteinstructure;
import java.io.BufferedReader;
import java.io.File;
import java.io.FileNotFoundException;
import java.io.FileOutputStream;
import java.io.FileReader;
import java.io.PrintStream;
import java.io.IOException;
import java.sql.ResultSet;
import java.sql.SQLException;
import java.sql.Statement;
import java.util.ArrayList;
import java.util.TreeMap;
import tools.MySQLConnection;


public class Graph {

        public final static String MYSQLSERVER="white";
        public final static String MYSQLUSER=getUserName();
        public final static String MYSQLPWD="nieve";

        ArrayList<Contact> contacts;
        // nodes is a TreeMap of residue serials to residue types (3 letter code)
        TreeMap<Integer,String> nodes;
        String sequence;
        public String accode;
        public String chain;
        double cutoff;
        String ct;
        boolean directed=false;
        
        // these 2 fields only used when reading from db
        int graphid=0;
        int sm_id=0;
        
        /**
         * Constructs Graph object by passing ArrayList with contacts and TreeMap with nodes (res serials and types)
         * Must also pass contact type, cutoff, accession code and chain
         * @param contacts
         * @param nodes
         * @param sequence
         * @param cutoff
         * @param ct
         * @param accode
         * @param chain
         */
        public Graph (ArrayList<Contact> contacts, TreeMap<Integer,String> nodes, String sequence, double cutoff,String ct, String accode, String chain) {
                this.contacts=contacts;
                this.cutoff=cutoff;
                this.nodes=nodes;
                this.sequence=sequence;
                this.accode=accode;
                this.chain=chain;
                this.ct=ct;
                if (ct.contains("/")){
                        directed=true;
                }
        }
        
        /**
         * Constructs Graph object from graph db, given the dbname, accode, chaincode (classic pdb chain code), ct and cutoff
         * @param dbname
         * @param accode
         * @param chaincode
         * @param cutoff
         * @param ct
         */
        public Graph(String dbname, String accode, String chaincode, double cutoff, String ct) throws GraphIdNotFoundError{
                this.cutoff=cutoff;
                this.accode=accode;
                this.ct=ct;
                // we set the sequence to empty when we read from graph db. We don't have the full sequence in graph db
                // when we pass the sequence in getCM to the ContactMap constructor we want to have either a full sequence (with unobserveds) or a blank in case we don't have the info
                this.sequence=""; 
                //TODO graphs in db are never directed, so this doesn't really apply here. Must solve all this!
                if (ct.contains("/")){
                        directed=true;
                }
                MySQLConnection conn = new MySQLConnection(MYSQLSERVER,MYSQLUSER,MYSQLPWD,dbname);
                getgraphid(conn, chaincode); // initialises graphid, sm_id and chain
                read_graph_from_db(conn); // gets contacts, nodes and sequence
                conn.close();
        }

        /**
         * Constructs Graph object by reading a file with contacts
         * An object created with this constructor will be missing the fields sequence and nodes
         * That means it's not possible to get a ContactMap from it using getCM because CM needs both sequence and nodes
         * @param contactsfile
         * @param cutoff
         * @param ct
         * @throws IOException
         * @throws FileNotFoundException
         */
        public Graph (String contactsfile, double cutoff,String ct) throws IOException, FileNotFoundException{
                this.cutoff=cutoff;
                this.ct=ct;
                // we set the sequence to blank when we read from file as we don't have the full sequence
                this.sequence="";
                if (ct.contains("/")){
                        directed=true;
                }
                read_contacts_from_file(contactsfile);
        }

        //TODO implement (from python) write_graph_to_db, do we really need it here??

        /** get user name from operating system (for use as database username) */
        private static String getUserName() {
                String user = null;
                user = System.getProperty("user.name");
                if(user == null) {
                        System.err.println("Could not get user name from operating system. Exiting");
                        System.exit(1);
                }
                return user;
        }

        public void read_contacts_from_file (String contactsfile) throws FileNotFoundException, IOException {
                //TODO eventually should read also nodes: either from file or obtain them from contacts (that would give only anyway nodes with contacts)
                contacts = new ArrayList<Contact>();
                System.out.println("Reading contacts from file "+contactsfile);
                BufferedReader fcont = new BufferedReader(new FileReader(new File(contactsfile)));
                String line;
                while ((line = fcont.readLine() ) != null ) {
                        int i = Integer.parseInt(line.split("\\s+")[0]);
                        int j = Integer.parseInt(line.split("\\s+")[1]);
                        contacts.add(new Contact(i,j));
                }
                fcont.close();
        }
        
        /**
         * Reads contacts and nodes from db.
         * The db must be a graph db following our standard format, i.e. must have tables: 
         * chain_graph, single_model_graph, single_model_node, single_model_edge
         * We don't care here about the origin of the data (msdsd, pdbase, predicted) for the generation of the graph as long as it follows our data format
         * We read both edges and nodes from single_model_edge and single_model_node.
         * The sequence is taken from nodes, thus it won't have unobserved or non standard aas.
         * @param conn
         */
        public void read_graph_from_db(MySQLConnection conn){
                contacts = new ArrayList<Contact>();
                nodes = new TreeMap<Integer, String>();
                try {
                        // we read only half of the matrix (contacts in one direction only) so that we have the same type of contacts as when creating Graph from Pdb object
                        String sql="SELECT i_num,j_num FROM single_model_edge WHERE graph_id="+graphid+" AND j_num>i_num ORDER BY i_num,j_num ";
                        Statement stmt = conn.createStatement();
                        ResultSet rsst = stmt.executeQuery(sql);
                        while (rsst.next()) {
                                int i=rsst.getInt(1);
                                int j=rsst.getInt(2);
                                contacts.add(new Contact(i,j));
                        }
                        rsst.close();
                        stmt.close();
                        sql="SELECT num,res FROM single_model_node WHERE graph_id="+graphid+" ORDER BY num ";
                        stmt = conn.createStatement();
                        rsst = stmt.executeQuery(sql);
                        while (rsst.next()){
                                int num=rsst.getInt(1);
                                String res=rsst.getString(2);
                                nodes.put(num, AA.oneletter2threeletter(res));
                        }
                        rsst.close();
                        stmt.close();
                } catch (SQLException e) {
                        e.printStackTrace();
                }

        }
        
        public void getgraphid (MySQLConnection conn, String chaincode) throws GraphIdNotFoundError{
                // input is chaincode i.e. pdb chain code
        // we take chain (internal chain identifier, pchain_code for msdsd and asym_id for pdbase) from pchain_code field in chain_graph 
        // (in the chain_graph table the internal chain identifier is called 'pchain_code')
                int pgraphid=0;
                String chainstr="='"+chaincode+"' ";
                if (chaincode.equals("NULL")){
                        chainstr=" IS NULL ";
                }
                try {
                        String sql="SELECT graph_id, pchain_code FROM chain_graph WHERE accession_code='"+accode+"' AND chain_pdb_code"+chainstr+" AND dist="+cutoff;
                        Statement stmt = conn.createStatement();
                        ResultSet rsst = stmt.executeQuery(sql);
                        int check=0;
                        while (rsst.next()) {
                                check++;
                                pgraphid=rsst.getInt(1);
                                chain=rsst.getString(2);
                        }
                        if (check!=1){
                                System.err.println("No pgraph_id match or more than 1 match for accession_code="+accode+", chain_pdb_code="+chaincode+", dist="+cutoff);
                        }
                        rsst.close();
                        stmt.close();
                        // we set the ctstr to the same as ct except in ALL case, where it is BB+SC+BB/SC
                        String ctstr=ct;
                        if (ct.equals("ALL")){
                                ctstr="BB+SC+BB/SC";
                        }
                        sql="SELECT graph_id,single_model_id FROM single_model_graph WHERE pgraph_id="+pgraphid+" AND CT='"+ctstr+"' AND dist="+cutoff+" AND CR='(true)' AND CW=1";
                        stmt = conn.createStatement();
                        rsst = stmt.executeQuery(sql);
                        check=0;
                        while (rsst.next()){
                                check++;
                                graphid=rsst.getInt(1);
                                sm_id=rsst.getInt(2);
                        }
                        if (check!=1){
                                System.err.println("No graph_id match or more than 1 match for pgraph_id="+pgraphid+", CT="+ctstr+" and cutoff="+cutoff);
                                throw new GraphIdNotFoundError("No graph_id match or more than 1 match for pgraph_id="+pgraphid+", CT="+ctstr+" and cutoff="+cutoff);
                        }
                } catch (SQLException e) {
                        e.printStackTrace();
                }
                
        }
        
        public void write_contacts_to_file (String outfile) throws IOException {
                PrintStream Out = new PrintStream(new FileOutputStream(outfile));
                for (Contact pair:contacts){
                        int i_resser=pair.i;
                        int j_resser=pair.j;
                        Out.println(i_resser+"\t"+j_resser);
                }
                Out.close();            
        }
        
        public ContactMap getCM() {
                // residues is the map from residue nums to residue types used in ContactMap class, i.e. it is the same as Pdb.resser2restype or Graph.nodes
                TreeMap<Integer,String> residues = new TreeMap<Integer,String>();
                // we copy residues from nodes (deep copy) 
                for (int node:nodes.keySet()){
                        residues.put(node, nodes.get(node));
                }
                // check if we are in directed or undirected case. If undirected we fill the opposite contacts to pass a full list of contacts to ContactMap (which contains full matrix)
                ArrayList<Contact> contacts2pass = new ArrayList<Contact>();
                if (directed){
                        contacts2pass=contacts;
                } else {
                        for (Contact cont:contacts){
                                int i_resser = cont.i;
                                int j_resser = cont.j;
                                contacts2pass.add(new Contact(i_resser,j_resser));
                                contacts2pass.add(new Contact(j_resser,i_resser));
                        }
                }
                // construct the ContactMap object and return it
                ContactMap cm = new ContactMap(contacts2pass,residues,sequence);
                return cm;
                
        }
        
}
Revision:	143
Committed:	Tue May 15 10:06:47 2007 UTC (17 years, 4 months ago) by duarte
File size:	9215 byte(s)
Log Message:	FIXED BUG: last commit didn't fix bug correctly: Changed the ContactMap constructor to put into residueNums and residueTypes nums from 1 to maximum of residues.keySet() Made public accode and chain fields in Graph
Line	User	Rev	File contents
1	duarte	123	package proteinstructure;
2	duarte	134	import java.io.BufferedReader;
3			import java.io.File;
4			import java.io.FileNotFoundException;
5	duarte	123	import java.io.FileOutputStream;
6	duarte	134	import java.io.FileReader;
7	duarte	123	import java.io.PrintStream;
8			import java.io.IOException;
9	duarte	135	import java.sql.ResultSet;
10			import java.sql.SQLException;
11			import java.sql.Statement;
12	duarte	123	import java.util.ArrayList;
13	duarte	129	import java.util.TreeMap;
14	duarte	135	import tools.MySQLConnection;
15	duarte	123
16
17			public class Graph {
18
19	duarte	135	public final static String MYSQLSERVER="white";
20			public final static String MYSQLUSER=getUserName();
21			public final static String MYSQLPWD="nieve";
22
23	duarte	123	ArrayList<Contact> contacts;
24	duarte	135	// nodes is a TreeMap of residue serials to residue types (3 letter code)
25	duarte	129	TreeMap<Integer,String> nodes;
26			String sequence;
27	duarte	143	public String accode;
28			public String chain;
29	duarte	123	double cutoff;
30			String ct;
31	duarte	135	boolean directed=false;
32	duarte	123
33	duarte	135	// these 2 fields only used when reading from db
34			int graphid=0;
35			int sm_id=0;
36
37	duarte	134	/**
38			* Constructs Graph object by passing ArrayList with contacts and TreeMap with nodes (res serials and types)
39			* Must also pass contact type, cutoff, accession code and chain
40			* @param contacts
41			* @param nodes
42			* @param sequence
43			* @param cutoff
44			* @param ct
45			* @param accode
46			* @param chain
47			*/
48	duarte	129	public Graph (ArrayList<Contact> contacts, TreeMap<Integer,String> nodes, String sequence, double cutoff,String ct, String accode, String chain) {
49	duarte	123	this.contacts=contacts;
50			this.cutoff=cutoff;
51	duarte	129	this.nodes=nodes;
52			this.sequence=sequence;
53			this.accode=accode;
54			this.chain=chain;
55	duarte	123	this.ct=ct;
56	duarte	129	if (ct.contains("/")){
57			directed=true;
58			}
59	duarte	123	}
60	duarte	135
61			/**
62	duarte	141	* Constructs Graph object from graph db, given the dbname, accode, chaincode (classic pdb chain code), ct and cutoff
63	duarte	135	* @param dbname
64			* @param accode
65	duarte	141	* @param chaincode
66	duarte	135	* @param cutoff
67			* @param ct
68			*/
69	duarte	141	public Graph(String dbname, String accode, String chaincode, double cutoff, String ct) throws GraphIdNotFoundError{
70	duarte	135	this.cutoff=cutoff;
71			this.accode=accode;
72			this.ct=ct;
73	duarte	142	// we set the sequence to empty when we read from graph db. We don't have the full sequence in graph db
74			// when we pass the sequence in getCM to the ContactMap constructor we want to have either a full sequence (with unobserveds) or a blank in case we don't have the info
75			this.sequence="";
76	duarte	135	//TODO graphs in db are never directed, so this doesn't really apply here. Must solve all this!
77			if (ct.contains("/")){
78			directed=true;
79			}
80			MySQLConnection conn = new MySQLConnection(MYSQLSERVER,MYSQLUSER,MYSQLPWD,dbname);
81	duarte	141	getgraphid(conn, chaincode); // initialises graphid, sm_id and chain
82			read_graph_from_db(conn); // gets contacts, nodes and sequence
83	duarte	135	conn.close();
84			}
85	duarte	129
86	duarte	134	/**
87			* Constructs Graph object by reading a file with contacts
88			* An object created with this constructor will be missing the fields sequence and nodes
89			* That means it's not possible to get a ContactMap from it using getCM because CM needs both sequence and nodes
90			* @param contactsfile
91			* @param cutoff
92			* @param ct
93			* @throws IOException
94			* @throws FileNotFoundException
95			*/
96			public Graph (String contactsfile, double cutoff,String ct) throws IOException, FileNotFoundException{
97			this.cutoff=cutoff;
98			this.ct=ct;
99	duarte	142	// we set the sequence to blank when we read from file as we don't have the full sequence
100			this.sequence="";
101	duarte	134	if (ct.contains("/")){
102			directed=true;
103			}
104			read_contacts_from_file(contactsfile);
105			}
106
107	duarte	135	//TODO implement (from python) write_graph_to_db, do we really need it here??
108
109			/** get user name from operating system (for use as database username) */
110			private static String getUserName() {
111			String user = null;
112			user = System.getProperty("user.name");
113			if(user == null) {
114			System.err.println("Could not get user name from operating system. Exiting");
115			System.exit(1);
116			}
117			return user;
118			}
119
120	duarte	134	public void read_contacts_from_file (String contactsfile) throws FileNotFoundException, IOException {
121	duarte	142	//TODO eventually should read also nodes: either from file or obtain them from contacts (that would give only anyway nodes with contacts)
122	duarte	134	contacts = new ArrayList<Contact>();
123			System.out.println("Reading contacts from file "+contactsfile);
124			BufferedReader fcont = new BufferedReader(new FileReader(new File(contactsfile)));
125			String line;
126			while ((line = fcont.readLine() ) != null ) {
127			int i = Integer.parseInt(line.split("\\s+")[0]);
128			int j = Integer.parseInt(line.split("\\s+")[1]);
129			contacts.add(new Contact(i,j));
130			}
131			fcont.close();
132			}
133
134	duarte	135	/**
135			* Reads contacts and nodes from db.
136			* The db must be a graph db following our standard format, i.e. must have tables:
137			* chain_graph, single_model_graph, single_model_node, single_model_edge
138			* We don't care here about the origin of the data (msdsd, pdbase, predicted) for the generation of the graph as long as it follows our data format
139			* We read both edges and nodes from single_model_edge and single_model_node.
140			* The sequence is taken from nodes, thus it won't have unobserved or non standard aas.
141			* @param conn
142			*/
143			public void read_graph_from_db(MySQLConnection conn){
144			contacts = new ArrayList<Contact>();
145			nodes = new TreeMap<Integer, String>();
146			try {
147	duarte	142	// we read only half of the matrix (contacts in one direction only) so that we have the same type of contacts as when creating Graph from Pdb object
148	duarte	135	String sql="SELECT i_num,j_num FROM single_model_edge WHERE graph_id="+graphid+" AND j_num>i_num ORDER BY i_num,j_num ";
149			Statement stmt = conn.createStatement();
150			ResultSet rsst = stmt.executeQuery(sql);
151			while (rsst.next()) {
152			int i=rsst.getInt(1);
153			int j=rsst.getInt(2);
154			contacts.add(new Contact(i,j));
155			}
156			rsst.close();
157			stmt.close();
158			sql="SELECT num,res FROM single_model_node WHERE graph_id="+graphid+" ORDER BY num ";
159			stmt = conn.createStatement();
160			rsst = stmt.executeQuery(sql);
161			while (rsst.next()){
162			int num=rsst.getInt(1);
163			String res=rsst.getString(2);
164			nodes.put(num, AA.oneletter2threeletter(res));
165			}
166			rsst.close();
167			stmt.close();
168			} catch (SQLException e) {
169			e.printStackTrace();
170			}
171
172			}
173
174	duarte	141	public void getgraphid (MySQLConnection conn, String chaincode) throws GraphIdNotFoundError{
175			// input is chaincode i.e. pdb chain code
176			// we take chain (internal chain identifier, pchain_code for msdsd and asym_id for pdbase) from pchain_code field in chain_graph
177			// (in the chain_graph table the internal chain identifier is called 'pchain_code')
178	duarte	135	int pgraphid=0;
179	duarte	141	String chainstr="='"+chaincode+"' ";
180			if (chaincode.equals("NULL")){
181			chainstr=" IS NULL ";
182			}
183	duarte	135	try {
184	duarte	141	String sql="SELECT graph_id, pchain_code FROM chain_graph WHERE accession_code='"+accode+"' AND chain_pdb_code"+chainstr+" AND dist="+cutoff;
185	duarte	135	Statement stmt = conn.createStatement();
186			ResultSet rsst = stmt.executeQuery(sql);
187			int check=0;
188			while (rsst.next()) {
189			check++;
190			pgraphid=rsst.getInt(1);
191	duarte	141	chain=rsst.getString(2);
192	duarte	135	}
193			if (check!=1){
194	duarte	141	System.err.println("No pgraph_id match or more than 1 match for accession_code="+accode+", chain_pdb_code="+chaincode+", dist="+cutoff);
195	duarte	135	}
196			rsst.close();
197			stmt.close();
198			// we set the ctstr to the same as ct except in ALL case, where it is BB+SC+BB/SC
199			String ctstr=ct;
200			if (ct.equals("ALL")){
201			ctstr="BB+SC+BB/SC";
202			}
203			sql="SELECT graph_id,single_model_id FROM single_model_graph WHERE pgraph_id="+pgraphid+" AND CT='"+ctstr+"' AND dist="+cutoff+" AND CR='(true)' AND CW=1";
204			stmt = conn.createStatement();
205			rsst = stmt.executeQuery(sql);
206			check=0;
207			while (rsst.next()){
208			check++;
209			graphid=rsst.getInt(1);
210			sm_id=rsst.getInt(2);
211			}
212			if (check!=1){
213			System.err.println("No graph_id match or more than 1 match for pgraph_id="+pgraphid+", CT="+ctstr+" and cutoff="+cutoff);
214			throw new GraphIdNotFoundError("No graph_id match or more than 1 match for pgraph_id="+pgraphid+", CT="+ctstr+" and cutoff="+cutoff);
215			}
216			} catch (SQLException e) {
217			e.printStackTrace();
218			}
219
220			}
221
222	duarte	123	public void write_contacts_to_file (String outfile) throws IOException {
223			PrintStream Out = new PrintStream(new FileOutputStream(outfile));
224			for (Contact pair:contacts){
225			int i_resser=pair.i;
226			int j_resser=pair.j;
227			Out.println(i_resser+"\t"+j_resser);
228			}
229			Out.close();
230			}
231	duarte	129
232			public ContactMap getCM() {
233			// residues is the map from residue nums to residue types used in ContactMap class, i.e. it is the same as Pdb.resser2restype or Graph.nodes
234			TreeMap<Integer,String> residues = new TreeMap<Integer,String>();
235			// we copy residues from nodes (deep copy)
236			for (int node:nodes.keySet()){
237			residues.put(node, nodes.get(node));
238			}
239			// check if we are in directed or undirected case. If undirected we fill the opposite contacts to pass a full list of contacts to ContactMap (which contains full matrix)
240			ArrayList<Contact> contacts2pass = new ArrayList<Contact>();
241			if (directed){
242			contacts2pass=contacts;
243			} else {
244			for (Contact cont:contacts){
245			int i_resser = cont.i;
246			int j_resser = cont.j;
247			contacts2pass.add(new Contact(i_resser,j_resser));
248			contacts2pass.add(new Contact(j_resser,i_resser));
249			}
250			}
251			// construct the ContactMap object and return it
252			ContactMap cm = new ContactMap(contacts2pass,residues,sequence);
253			return cm;
254
255			}
256
257	duarte	123	}